An analysis of the Australian Constitutional Referendum Study
by John Pyke
(Lecturer in law, Queensland University of Technology
The "vote no to this republic" campaigners who really wanted a DE republic were nearer the mark - almost 55% of the people voted their way - but even they, as I show below, were more dogmatic than many of their followers. But what did the people really really want?
There were a lot of newspaper polls in the 1990s that asked maybe 600 to 2000 people what their first choice was out of the 2 most common republic models or retaining the monarchy, but first preferences alone are not terribly enlightening. Someone may prefer a republic with the President directly elected by the people (a Direct Election, or "DE" republic, in my terminology), but be happy enough with a President selected by Parliament (a Parliamentary Selection, or "PS" republic), whereas other supporters of DE may find PS so appalling that they would rather keep the Queen, if all that is being offered is a PS republic. To understand what the people really really want you must ask them about their first and second preferences, and you must know what first preference was held by each of the people expressing a second preference.
The most authoritative information on what the people of Australia really really wanted in late 1999, just after the Republic referendum, is provided by the Australian Constitutional Referendum Study, a survey conducted by David Gow (U of Q), Clive Bean (QUT) and Ian McAllister (ANU). They sent 6150 survey forms to electors chosen from the electoral roll and received 3431 back (over 500 from each State, 57 from the ACT, 22 from the NT, and 28 who failed to say what area they lived in!). The results are held on the Social Science Data Archives at the ANU (you need to register as a user to see the data, but it's not a difficult process). This is a far bigger and therefore more accurate sample of peoples' opinions than any of the polls published in the newspapers at the time of the referendum. Because of the number and nature of the questions asked, it was possible to do some checks as to the probable accuracy of the figures. This is discussed in the Notes below - it will suffice to say here that the responses may have overstated the general support for a republic by 2%, but no more. Of course the Survey is now (early 2007) 6 1/2 years old, but I know of no evidence of general shifts in the population's views about a republic - we certainly do not seem to be getting any more attached to the British monarchy as the years go by.
I should start with a table of the answers to the "first preference" question. Here it is:
| Table 1: First preferences | DE republic | PS republic | Keep Queen | Total with an opinion | Don't know | No answer | Total |
|---|---|---|---|---|---|---|---|
| Number | 1719 | 637 | 747 | 3103 | 88 | 240 | 3431 |
| Percentage of those with an opinion | 55.4 | 20.5 | 24.1 | 100 | - | - | - |
| Percentage of all respondents | 50.1 | 18.6 | 21.8 | 90.4 | 2.6 | 7.0 | 100 |
The useful thing about the Survey is that it asked respondents to state their first preference and second preference as between a DE republic, a PS republic, and retaining the monarchy. Of course by themselves the 2nd-preference answers mean nothing - for example, the number who had a DE republic as second preference helps you understand nothing unless you know how many of them really wanted, as their first preference, a PS republic and how many wanted to keep the monarchy. But the Archives include an analysis tool so you can do clever things like cross-tabulation of answers or "recoding" to change the form of presentation of the tables. When I did the analysis below, the program was the Online Analysis System In Social Sciences (OASISS), but it has now been replaced by something called NESSTAR. (I haven't used it yet but I gather it produces something more like an MS Excel file.) The figures shown below are the result of my analysis of the results, and neither Gow, Bean and McAllister nor the Social Science Data Archives bear any responsibility for them.
So Table 2, below, shows the distribution of the first preferences and second preferences. Percentages are expressed out of the 3103 respondents to the survey who stated at least a first preference (and, I guess, are likely to bother voting in a referendum). The figures in all columns will be a bit lower if you calculate perecentages out of those who gave an answer including "don't know" (3191) or all of those who returned a survey form (3431).
For the "or NO change!" columns, I have added those who obstinately repeated their first preference as their second, those who answered "Don't Know" and those who gave no second preference, to those who expressly said "Keep the Queen" as their second preference. [See the more detailed figures from which I compiled this table in the Notes.] The "Don't Knows" (10% of those who had recorded a first preference) may still have been waiting to make up their mind, but I assume those who repeated their first preference or would not answer the question were impliedly saying "give me my first choice - I'm not interested in anything else". The proportion who did not even bother to answer the second preference question was particularly high (6%) among DE supporters - understandable, because they had been so alienated by the process leading up to the referendum.
| Table 2: First and second preferences | 1. DE, or NO change! | 2. DE, with PS 2nd pref | 3. PS, with DE 2nd pref | 4. PS, or NO change! | 5. Keep Queen, DE 2nd pref | 6. Keep Queen, PS 2nd pref | 7. Keep Queen, no 2nd pref |
|---|---|---|---|---|---|---|---|
| Number | 965 | 754 | 484 | 153 | 480 | 118 | 149 |
| Percentage (out of the 3103 with a 1st pref) | 31.1 | 24.3 | 15.6 | 5.0 | 15.5 | 3.8 | 4.8 |
| Prepared to accept PS? (Cols 1+2+3) | No! | 44.8 | No! | (No, but PS best if we must have a republic) | No! | ||
| Actual "yes" vote in 1999 | - | 45.1 | - | - | - | ||
| Prepared to accept DE? (Cols 2+3+4) | 71.0 | No! | (No, but DE best if we must have a republic) | No! | No! | ||
You can see (contrary to the evidence mentioned above that the survey overstates the number of republicans) that the figures in the first line actually slightly understate the "yes" vote compared to that received by the real PS republic proposed in 1999 (45.1%). But the significant thing is that the numbers are close - so close that it must be a bit fluky, even given the large sample size. But it clearly shows that people did, by and large, vote as you would expect from the expression of first and second preferences in the survey.
Therefore I think we can be pretty confident that the bottom line is a good predictor of how people will vote if a DE republic is proposed sometime - it will be supported by between 65 and 70% of the people of Australia. That is, what the people really, really want is a republic, and the only republic that a majority of them will be prepared to support is one where We the People have a direct part in the election of a President.
Looking at Table 2 again, we can see that, unlike their self-appointed leaders, most people who preferred a Parliamentary Selection republic were not dogmatic about it and will be reasonably happy to accept a DE republic when it is eventually offered (as it must be!). Out of those whose first preference was a PS republic, only 24% (counting, as already noted the "no 2nd pref" people as implied "PS or nothing" people) shared their self-appointed "leaders'" "PS republic or nothing" approach. This amounts to only 5% of the population. Those who tried to whip up support for a "yes" vote by attacking the alternative concept of a republic should realise that their dogmatic view is shared by, at most, a mere 5% of the people of Australia!
On the other hand, the direct electionists who took over the effective leadership of the "no" campaign were somewhat more in touch with their followers - their "Direct Election or nothing" approach is indeed shared by 56% of ordinary people who would prefer a DE republic. Even in this group, though, the other 44% are flexible enough (and were flexible enough, in 1999) to accept what is for them a second-best republic. [Which was just as well for the "yes" campaigners - the "yes" vote would have been truly pathetic if not for the effective second-preference votes of DE republicans - the 45% who voted "yes" consisted of about 20% who really wanted a PS republic and 25% who were prepared (even if grudgingly) to accept it as second best.]
Ther survey doesn't tell us anything about the degree of intensity and dogmatism among monarchists (ordinary monarchists, that is, compared to their self-appointed leaders) - presumably some will be devasted if we abandon the monarchy for any republic, while others will quickly learn to love the new republic. (Perhaps the 17 of them (see the table in the Notes) who gave "Retain Queen" as both their first and second preference were saying "give me the Queen or give me death", but this is the wildest speculation on my part!) The significant thing here is that even among monarchists, opposition to a republic is likely to be less passionate if a DE republic is offered next time. Compared to the too-clever-by-half monarchists who recommended a "yes" vote because we might be offered the "even worse" alternative next time, it is clear that two-thirds of monarchists believe that if they must have a republic, they would prefer to go all the way with democracy and have one where the people elect the President. Though I have not tabulated the monarchists' second preferences in the tables below, this in fact applies equally strongly among monarchists from all States, and regardless of what political party they support.
First, it may be useful to look at how survey respondents' first preferences varied by State or Territory. This is shown in Table 3. As the samples are only about 500 from each state, these figures could misrepresent the true population by up to 10% (though two-thirds of the figures should have an error of less than 5%), and I really shouldn't be quoting tenths of percents! With due allowance for this, the table shows that though there were some variations by State they are not huge - all across the country, a majority want a DE republic. Figures that show a State had a particularly high or low preference for an option (on an impressionistic basis, not by significance testing) are shown in bold or are underlined respectively. The States are arranged in increasing order of support for PS as you read down; the Territories are left at the bottom because of their small sample size, but the ACT fits neatly after Victoria on the PS-support criterion anyway.
| Table 3: 1st Preferences by State | Pres-direct election (No; % of opinions) | Pres-by Parliament (No; % of opinions) | Retain Queen (No; % of opinions) | Total with an opinion | dk or na* (No; % of total) | Total |
|---|---|---|---|---|---|---|
| Queensland | 265; 57.0% | 61; 13.1% | 139; 29.9% | 465 | 52; 10.1% | 517 |
| South Australia | 315; 57.5% | 101; 18.4% | 132; 24.1% | 548 | 59; 9.7% | 607 |
| New South Wales | 260; 57.0% | 93; 20.4% | 103; 22.6% | 456 | 51; 10.1% | 507 |
| Tasmania | 293; 55.4% | 112; 21.2% | 124; 23.4% | 529 | 59; 10.0% | 588 |
| Western Australia | 282; 56.2% | 117; 23.3% | 103; 20.5% | 502 | 46; 8.4% | 548 |
| Victoria | 253; 50.2% | 125; 24.8% | 126; 25.0% | 504 | 53; 9.5% | 557 |
| Aust Capital Terr | 28; 49.1% | 19; 33.3% | 10; 17.5% | 57 | 0 | 57 |
| Northern Territory | 13; 65% | 4; 20% | 3; 15% | 20 | 2; 9.1% | 22 |
| No State given | 10; 45.5% | 5; 22.7% | 7; 31.8% | 22 | 6; 21.4% | 28 |
| Total; avg. % | 1719; 55.4% | 637; 20.5% | 747; 24.1% | 3103 | 328; 9.6% | 3431 |
These figures confirm a couple of popular prejudices and negate a couple of others. There are proportionally more monarchists in Queensland - but that was not the reason for the low "yes" vote in Queensland at all. That was simply because there are very few PS republicans, and about the normal proportion of "DE or nothings" (with particularly vigorous leadership) among the DE republicans, in the State - most Queenslanders were saying, like most of everyone else, "give us a republic, but let us vote directly for the President". In fact, the proportion of DE supporters is about equal in all States except Victoria. The main variation is in the split between monarchy and PS; as one goes up, the other goes down, except for Victoria which has relatively high support for both at the expense of DE. And contrary to much imaginative speculation before the referendum, there are no great nests of monarchists among the British migrants in South or Western Australia (or if there are, they're swamped by the rest of the population).
There is no majority of PS republicans anywhere (outside of the Federal Parliament and possibly the membership of the ARM), but the federal politicans' proposal was relatively more favoured in the ACT, the centre of federal politics (where it gained a majority in the referendum, but only with the help of DE republicans who voted "yes" in the absence of anything better), followed by Victoria (where, as I note below, there are more dogmatic "PS or nothing" republicans).
As to the careless ones who neglected to state their State, the high proportion of monarchists and "don't knows" or "no opinions" may confirm the popular suspicion that the populations of monarchists and the old and doddery overlap, but the low proportion of DE republicans refutes another popular tale, that DE republicans are less educated and/or intelligent. It seems, at least on this scant evidence, that DE republicans were more likely to fill in the survey carefully and completely than either of the other groups.
| Table 4: First and second preferences, by State or Territory | No. with a 1st preference | 1. DE, or NO change! % | 2. DE, with PS 2nd pref % | 3. PS, with DE 2nd pref % | 4. PS, or NO change! % | Should vote for PS (2+3+4) % | Actual yes vote in 1999 % | Should vote for DE (1+2+3) % |
|---|---|---|---|---|---|---|---|---|
| Queensland | 465 | 34.4 | 22.6 | 9.0 | 4.1 | 35.7 | 37.4 | 66.0 |
| Tasmania | 549 | 31.4 | 24.0 | 16.6 | 4.5 | 44.1 | 40.4 | 72.0 |
| Western Australia | 502 | 29.5 | 26.7 | 19.5 | 3.8 | 50.0 | 41.5 | 75.7 |
| South Australia | 548 | 35.6 | 21.9 | 14.8 | 3.6 | 40.3 | 43.6 | 72.3 |
| New South Wales | 456 | 32.9 | 24.1 | 15.4 | 5.0 | 44.5 | 46.4 | 72.4 |
| Victoria | 504 | 23.8 | 26.4 | 17.1 | 7.7 | 51.2 | 49.8 | 67.3 |
| Aust Capital Territory | 57 | 21.1 | 28.1 | 22.8 | 10.5 | 61.4 | 63.3 | 72.0 |
| Northern Territory | 20 | 35 | 30 | 15 | 5 | 50 | 48.8 | 80 |
| No state given | 22 | figures vague, but still suggest a defeat for PS (36%) and support for DE (59%) | ||||||
| Total, or av% | 3103 | 31.1 | 24.3 | 15.6 | 4.9 | 44.8 | 45.1 | 71.0 |
The first thing to note is that, judged by the comparison of the predicted support for a PS republic and the actual vote in 1999, the samples seem to be fairly representative of actual opinion in all States except Western Australia, and even in both Territories. You can see that a DE proposal should pass in all States and Territories. It may pass by a slightly smaller margin in Queensland, where there are a few more monarchists and fewer PS republicans, and in Victoria, where it shows there are more "PS or nothing" republicans. [Of course, as the responses from Western Australia overestimate the number who supported a PS republic by 8% or 9%, they may, or may not, also overestimate those who will support the other republican proposal.] But it is really hard to imagine that the "yes" vote in any State or Territory could be much less than 60%, and it will be around 70% in most States and Territories. The "strain on federation" argument is a red herring - the people really, really want a republic in which they elect the President directly, and they want that all across Australia!
What's my evidence for this? Another cross-correlation of the Survey results. The Survey asked respondents to state their "party identification", which included "no party" for the swinging voters. [Respondents of course could also just ignore the question - where they ignored this question but answer the first and second preference question I've managed to count them and have given them their own line in the table.] Of course the figures are now 7 years old, and the party identification figures may have drifted a bit since then (2 parties, indeed, are fading out of sight), but I know of no evidence that the preferences of the general public on the republic issue have changed in any significant way. I actually ran the OASISS analysis some 6 years ago and stuck it in a drawer - I have just recently (early 2007) pulled it out and done the further manual layer of analysis to make sense of the numbers.
As with the States, above, it may be useful to first present a simple table of how first preferences correlate with party support. I have put the parties in decreasing order of tendency to monarchism, so that the patterns are easier to see. Here the percentages are out of those (totalling 3033) who gave an answer (even if it was "don't know") to both the first and second preference questions, so the overall percentages in the bottom line are different from either of the lines in Table 1 - but still very much in the same "ballpark".
| Table 5: First preferences by Party identification | No of respnts | DE 1st pref % | PS 1st pref % | Keep Queen % | Don't know (or care?) % |
|---|---|---|---|---|---|
| National | 84 | 45.2 | 11.9 | 41.7 | 1.2 |
| One Nation | 53 | 58.5 | 0 | 39.6 | 1.9 |
| No answer | 31 | 51.6> | 9.7 | 35.5 | 3.2 |
| Liberal | 1070 | 48.0 | 16.7 | 33.0 | 2.2 |
| Answered "No party" | 427 | 49.6 | 22.0 | 24.1 | 4.2 |
| Labor | 1149 | 59.4 | 23.8 | 14.7 | 2.0 |
| Aust Democrats | 118 | 55.1 | 31.4 | 11.9 | 1.7 |
| Greens | 91 | 54.9 | 34.1 | 4.4 | 6.6 |
| Other party | 10 | 90 | 10 | 0 | 0 |
| Total who answered & %'s of that | 3033 | 53.4 | 20.8 | 23.4 | 2.5 |
Interesting, eh - even surprising? There are differences between the followers of the different parties, but they are less than you might expect. Among supporters of the coalition and One Nation, the percentage who are monarchists is indeed higher than among those who support other parties, but monarchists are a definite minority even in the supporters of these parties. The interesting, and perhaps surprising, thing is that although support for DE is generally somewhat lower among the more-monarchical parties, it is relatively constant (52±7%) and that the more significant variation between parties is how the remainder of their supporters split between monarchy and PS. The 'right-wing' voters combine relatively high support for the monarchy with particularly low support of parliamentary selection of a President - the percentage in favour of PS is very low in absolute terms and also quite low as a percentage of the total republicans in these parties' samples.
Why should this be so? If you think of Labor/Green/Democrat voters as "radical" and Liberal/National/ON voters as "conservative" it seems paradoxical. But if you think of Labor/Green/Democrat voters as "collectivist" and Liberal/National/ON voters as "individualist" it all makes perfect sense. In the minds of many of the latter voters, the "conservative" instinct "don't change anything" lives side by side with the individualist instinct "don't let politicians make decisions for you". It seems that, among "conservative" voters on this issue, while a few manifest the first instinct, even more manifest the second one. Politicians who get elected to represent the Liberal or National parties may learn to love power, but those who have sent them to Canberra remain uneasy about it. (People who think politics is all about reason, and who mistrust instinct and emotions - read this with your right eye covered so it will go to your right brain and, the psychologists claim, you will 'understand' it better!)
Next, here is the analysis of the interaction of the first and second preferences by party identification. The parties are still sorted in order of decreasing support for the monarchy. Again the percentages are out of those who gave an answer (even if "don't know") to both the first and second preferences, so we may have lost track of some of those who stated a first preference but not a second one - and remember, 6% of those whose first preference was DE didn't bother to answer the second-preference question, so we've lost a few of the "DE or nothing" voters. But the results are still informative, possibly surprising, and conclusive.
| Table 6: First and second preferences, by party identification | 1. DE, or NO change! % | 2. DE, with PS 2nd pref % | 3. PS, with DE 2nd pref % | 4. PS, or NO change! % | Keep Queen % | Don't know (or care?) % |
|---|---|---|---|---|---|---|
| National | 33.3 | 11.9 | 10.7 | 1.2 | 41.7 | 1.2 |
| One Nation | 43.4 | 15.1 | 0 | 0 | 39.6 | 1.9 |
| No answer | 22.6 | 29.0 | 9.7 | 0 | 35.5 | 3.2 |
| Liberal | 27.1 | 20.9 | 11.5 | 5.2 | 33.0 | 2.2 |
| Answered "No party" | 31.4 | 18.3 | 15.5 | 6.5 | 24.1 | 4.2 |
| Labor | 28.3 | 31.1 | 20.2 | 3.6 | 14.7 | 2.0 |
| Aust Democrats | 22.9 | 32.2 | 21.2 | 10.2 | 11.9 | 1.7 |
| Greens | 25.3 | 29.7 | 27.5 | 6.6 | 4.4 | 6.6 |
| Other party | 70 | 20 | 10 | 0 | 0 | 0 |
| Total or overall % | 28.5 | 24.9 | 16.0 | 4.80 | 23.4 | 2.5 |
Now this may really surprise a lot of people (it rather surprised me, in fact!). There is no great concentration of "give me PS or we'll keep the Queen till Hell freezes over" people among those who vote for the conservative parties - whatever attitude party members and politicians might have - and none at all in the sample of 53 One Nation voters. What we do find among the conservative voters is the opposite - a strong concentration of "DE or no change" voters. Run your eyes down columns 1 and 2 - up among the conservative parties the figure in 1 is much bigger than the one in 2; by the time you get down to Labor and the Greens it's a bit smaller. The polarisation in the conservative voters between monarchy and DE continues - not only do the republicans among them prefer DE more strongly than average, but the DE-preferers among them (and among the swinging voters as well) are more strongly "DE or no change" than anybody else. The 1999 referendum was defeated because it was opposed by conservative voters - well, everybody knew that, you may say! But the figures from the above table make the reason more clear - some conservative voters voted "no" because they were monarchists, and almost as many voted "no" because they were "DE or no change" republicans. Clem Jones and Phil Cleary may have led the campaign for direct election in 1999, but many of those who were listening to them were from the other side of the political divide. [I haven't tabulated the monarchists' second preferences, but the raw data show that among the monarchists in every party, the great majority of them would prefer DE if we must have a republic.]
In the next table I show how these attitudes should reflect in the vote on a referendum proposing a change to a specific form of republic. The format is a bit different from any tables above: I thought in this table it may be useful to show how many would vote for each model because it is their really-preferred model, and how many would reluctantly accept it as second-best if it is the only republic on offer. [You can check the figures against Table 6 - for example the first column below is the sum of columns 3 and 4 above.] Remember, the State-by-State figures for voting for PS, above, are generally close to the actual vote, so the general reliability of the Survey as a predictor of how people felt about republican 'models', and how those feelings were converted into votes, seems confirmed - so the party-by-party figures should be accurate too.
| Table 7 | Would logically vote for PS... | Would logically vote for DE... | ||||
|---|---|---|---|---|---|---|
| Party | as 1st pref | as 2nd pref | Total | as 1st pref | as 2nd pref | Total |
| National | 11.9 | 11.9 | 23.8 | 45.2 | 10.7 | 55.9 |
| One Nation | 0 | 15.1 | 15.1 | 58.5 | 0 | 58.5 |
| not answered | 9.7 | 29.0 | 38.7 | 51.6 | 9.7 | 60.3 |
| Liberal | 16.7 | 20.9 | 37.6 | 48.0 | 11.5 | 59.5 |
| Answered "no party" | 22.0 | 18.3 | 40.3 | 49.6 | 15.5 | 65.1 |
| Labor | 23.8 | 31.1 | 54.9 | 59.4 | 20.2 | 79.6 |
| Aust Dems | 31.4 | 32.2 | 63.6 | 55.1 | 21.2 | 76.3 |
| Greens | 34.1 | 29.7 | 63.8 | 54.9 | 27.5 | 72.4 |
| "Other" | 10 | 20 | 30 | 90 | 10 | 100 |
| Overall % | 20.8 | 24.3 | 45.1 | 53.4 | 15.6 | 69.0 |
This is almost spooky, in that the "predicted" PS vote matches the actual "yes" vote in 1999 exactly. The predicted DE vote is 2% lower than in Tables 2 and 4, but don't forget that this analysis has omitted some of the DE voters who refused to give any answer as to second preference.
On the other hand the table shows that if a reasonable direct-election proposal is ever put to the people, it will be supported by a significant-to-substantial majority of voters for every party. The "PS or nothing" (also known as "DE is frightening") view is only subscribed to by about 5% of the general population, and about double that proportion among Democrat voters. (How droll - more fear of democracy in Democrats' supporters than in other parties!). It also seems, from my observations, to be particularly prevalent among members of the Commonwealth Parliament and members of the Australian Republican Movement. This view, or fear, is based not only on a considerable degree of ignorance of the way that real republican constitutions work in real republican countries, but also on a good dose of snobbery about hoi polloi, the people. Now admittedly (see above) some of the opposition to PS is also based on ignorance of the fact that it can work quite well elsewhere (both sides are prone to hysteria), but anti-DE snobs would like to think of themselves as better-educated - they should study the way that DE works elsewhere and get over their fears!
None of this is to say that Australians want dramatic changes to our institutions - another of the questions in the Survey was "should the President have more powers than the Governor-General, or less, or about the same?", and the great majority answered "about the same". Of PS supporters, only 10% thought the President should have more power, and 30% of DE supporters thought so (thanks to David Latimer for doing the analysis and passing the figures on). That is, most of us want a republic where the President does about the same job as the Queen in the UK, the President does in Iceland or Ireland, or the Governor-General does here at present. That is, the President would formally appoint Ministers who had the support of the majority in the House of Representatives, and formally dismiss them when convention demanded that they should resign but they refused to do so. Apart from that s/he would routinely act on the advice of the Ministers, formally signing the appointments of Judges and other documents where the law currently requires the decision to be made by the "Governor-General in Council", and generally acting as a symbol of the nation's independence and democratic institutions. There is nothing scary about having an elected person do this, and the constitutional amendments are not hard to draft. The next Prime Minster - Labor or Liberal - should have it done, and should give us a chance to vote on the kind of republic that we know the majority wants.
People in politics or the ARM, or in both, who want to lead us to an Australian republic should reflect on this. The gut-feeling of the majority of Australians tells them that they want a republic, and that they want to vote for the President directly, and this is particularly strong among conservative voters. DE has a strong symbolic power - as one of my sons asked the other day "DE is a symbol of the sovereignty of the people; what exactly is PS symbolic of?" I could only give lame answers. Republican campaigners and politicians - conservative politicians in particular - who campaign against a directly-elected President risk alienating themselves from their supporters on an issue - the issue - of national symbolism and identity.
Notes on the accuracy of the Survey and how I compiled the tables - for those who want to double-check
Percentages of what?
In the survey leaders' own tables of answers, they have calculated percentages as a percentage of the people who recorded some "response" - even a "don't know" - to that particular question. For example, for the "first preference" question, there were 3191 answers including 88 "don't knows", and 240 "missing" (ie, blank) responses - so in the User's Guide the percentage for a DE republic is given as 1719/3191x100, which is 53.9%
Now I can't see any reason to treat those who expressly answered "don't know" differently from those who just left a question blank, so I count 3103 people as having answered this question, and the percentage of them for a DE republic is 55.4%. If we are to try to use the figures to predict how people will vote in a referendum, the behaviour of these "no preference" people is the big mystery - will they vote or not? Maybe they could be expected to act consistently, and either fail to vote or vote informal (they're about 10% of the population, and this corresponds pretty nearly with the number who either vote informal at elections or don't seem to vote at all). In that case, the 55.4% would be a good indicator of the real proportion prepared to vote in favour of this model (as their first preference), out of those who would bother to vote at all. But then perhaps those who have no real preference would take the "safe" course and vote "no"; in that case the percentage in favour of a change should be calculated out of the whole number of respondents, 3431 - which gives 50.1% for a DE republic. I have shown both figures in Table 1 - presumably the real percentage of those whose first preference is a DE republic, out of those who would bother to vote, is somewhere between the two (minus some part of the possible overcounting of republicans by 2 or 3 percent mentioned below). In the later tables I have followed my hunch that the "don't knows" will still be confused at a later referendum, and will either vote randomly or not at all, so in Tables 2-4 I have calculated percentages only out of those who expressed a first preference. In preparing Tables 5-7, OASISS would only analyse those who had expressed both a first and second preference, so that is what I have used.
Fortunately, one question was asked which allows us to check whether the returned forms were representative of the opinions of the people at large. Respondents were asked how they had voted in the republic referendum - the survey was taken within the 2 months after it, so this should have been fresh in their memory. This does reveal a slight systematic bias in the answers - "Yes" voters are overrepresented by 1.7 percent (46.8% instead of the actual result of 45.1%). Analysis by States shows that (judged on this basis) the figures from New South Wales are about right, and those from the other States and Territories put the "yes" vote too high by amounts ranging from 2.5% (South Australia) to 8% (Western Australia). It is also possible to work out how people should have voted, to be consistent with their expressed preferences, and to check this against actual voting - this has been mentioned above and seems to confirm the general representativeness of the sample from most States and Territories, and the fact that republicans were overrepresented in the WA sample.
By cross-tabulating other answers, you can also detect some carelessness and confusion (or flippancy? or being stoned?) among a small percentage of respondents. For example, of 88 who didn't seem to know what their first preference was, 57 managed to express a second preference, and 16 who said they were monarchists claimed to have voted "yes". Perhaps these can be explained as tactical votes on the ground "better vote for this republic, because if it fails we might be offered a worse one", but then six of them said as second preference they'd prefer a DE republic to the one that was on offer, so why would they have voted "yes"? And how do you explain the 106 people who said they really wanted a DE republic, with retention of the Queen as second preference, but who reported voting "yes"? Again, some of them may have had tactical considerations in mind, but it does all seem to add up to evidence that somewhere between one and, at most, three percent misunderstood the questions, either through dullness of mind or of eyesight, or treated the exercise less than seriously. There is no indication that these inconsistencies biased the overall results in any one direction more than another.
Another slight source of "error" is that roughly equal numbers of responses were received from each State, which would give the opinions held in the less-populous States too much weight in a simple count. In the "official" summaries in the Users' Guide, each table of answers is presented in a weighted and an unweighted version, but I did not correct for States' weighting in the Tables below, because when using OASISS it was difficult to do cross-tabulations and correct for weighting at the same time. However, in the case of the two questions about preferences for republic or monarchy the application of the weighting correction only decreased the figure for a republic and increased the figure for a monarchy by about half a percent, so the percentages given in all the tables should be correct within the same margin. Overall, then, the figures in the tables may overstate the number of republicans by two or three percent, but no more and possibly less. So a figure of 54% in favour of a particular proposal may be weakish evidence of majority support, but a figure of 70% - which we have found above - clearly is strong evidence!
| A14P2 (2nd preference)) | ||||||||
| A14P1 (1st pref) | Count | Pres-direct elec 1 | Pres-by Parlt 2 | Retain Queen 3 | Don't know 4 | Row Total | (1st pref, no 2nd) | (Real row total) |
|---|---|---|---|---|---|---|---|---|
| Pres-direct elect 1 | 27 | 754 | 666 | 171 | 1618 | 101 | 1719 | |
| Pres-by Parlt 2 | 484 | 6 | 101 | 38 | 629 | 8 | 637 | |
| Retain Queen 3 | 480 | 118 | 17 | 95 | 710 | 37 | 747 | |
| Don't know 4 | 29 | 10 | 18 | 19 | 76 | 12 | 88 | |
| Column Total | 1020 | 888 | 802 | 323 | 3033 | 158 | 3191 | |
| (No 1st pref) | .. | 240 | ||||||
| (Grand total) | .. | 3431 | ||||||
For the 3-way cross-correlations, OASISS produced a table like the above one for the figures for each State or for each party preference respectively. For the State by State figures I was able to follow a process as described above for each line of the table. For the party by party figures, as explained in the main text, it was difficult to find how many people had expressed a first preference but not a second one, so the "first preference or nothing" figures may be slightly low.